JUDGES HAVE WITHIN THEIR CAPABILITIES TODAY THE MEANS TO SHARPLY CURTAIL, IF NOT VIRTUALLY ERADICATE, SENTENCING DISPARITIES IN MOST AMERICAN JURISDICTIONS. THIS IS THE SINGLE CONCLUSION OF A 2-YEAR EFFORT TO DETERMINE THE FEASIBILITY OF LIMITING DISPARATE SENTENCES AT THE STATE COURT LEVEL BY PROVIDING SENTENCING GUIDELINES THAT STRUCTURE LIMITS FOR JUDICIAL DISCRETION. FOUR SITES WERE INVOLVED IN THE PROJECT--DENVER COUNTY, COLORADO, AND THE STATE OF VERMONT AS PARTICIPANTS, AND ESSEX COUNTY, NEW JERSEY, AND POLK COUNTY, IOWA, AS OBSERVERS. OVER 200 ITEMS OF INFORMATION FROM 200 RANDOMLY SELECTED SENTENCING DECISIONS IN EACH OF THE TWO PARTICIPATING COURTS WERE COLLECTED. THE INFORMATION WAS ANALYZED FOR OFFENSE/OFFENDER CHARACTERISTICS THAT STATISTICALLY ACCOUNTED FOR THE LARGEST PERCENTAGE OF VARIATION IN THE SENTENCING DECISION. IT WAS FOUND THAT OFFENSE SERIOUSNESS AND THE OFFENDER'S PRIOR CRIMINAL RECORD WERE THE TWO MOST INFLUENTIAL ITEMS OF INFORMATION IN THE SENTENCING DECISION. A SERIES OF GUIDELINE MODELS WERE DESIGNED AND TESTED AGAINST A SAMPLE OF ACTUAL CASES. THE MODELS THEN WERE SYNTHESIZED INTO ONE AND GUIDELINE SENTENCES OF THE MODEL WERE PRESENTED TO THE DENVER JUDICIARY FOR CONSIDERATION 2 TO 3 DAYS AFTER PRONOUNCING SENTENCE. THE JUDGES THEN PROVIDED THE RESEARCHERS WITH FEEDBACK EXPLAINING THE CAUSES OF DIFFERENCES IN THE ACTUAL SENTENCES FROM THE GUIDELINES SENTENCES WHEN THEY OCCURRED. MODEL SENTENCES WERE COMPUTED USING THE INFORMATION FROM THE JUDGES AND BY GIVING ASSIGNED WEIGHTS TO PARTICULAR AGGRAVATING AND MITIGATING FACTORS RELATING TO PERTINENT CHARACTERISTICS OF THE CRIME AND THE CRIMINAL. THE FINDINGS ALLOWED THE PROJECT TEAM TO CONCLUDE THAT IT IS FEASIBLE TO STRUCTURE JUDICIAL DISCRETION BY MEANS OF SENTENCING GUIDELINES.